test-static-http: flush access log per request It appears that stderr is fully buffered on Windows. # no-check-commit because of log_message() function

statichttprepo: do not use platform path separator to build a URL It wouldn't work between Windows client and Unix server.

merge: disable path conflict checking by default (issue5716) We shouldn't ship a severe perf regression in hg update for 4.4. Differential Revision: https://phab.mercurial-scm.org/D1223

merge: add a config option to disable path conflict checking We've found a severe perf regression in `hg update` caused by the path conflict checking code. The next patch will disable this by default. Differential Revision: https://phab.mercurial-scm.org/D1222

dirstate: clean up when restoring identical backups When a dirstate backup is restored, it is possible that no actual changes to the dirstate have been made. In this case, the backup is still a hardlink to the original dirstate. Unfortunately, `os.rename` silently fails (nothing happens, and no error occurs) when `src` and `dst` are hardlinks to the same file. As a result, the backup is left lying around. Over time, these files accumulate. When restoring dirstate backups, check if the backup and the dirstate are the same file, and if so, just delete the backup. Differential Revision: https://phab.mercurial-scm.org/D1201

tests: add a test demonstrating failure to clean up dirstate backups Differential Revision: https://phab.mercurial-scm.org/D1200

tests: adjust hooks for Windows I'm not sure why these weren't working on Windows. The failures were generally in the style of: - remote: phase-move: cb9a9f314b8b07ba71012fcdbc544b5a4d82ff5b: 1 -> 0 + remote: "phase-move: $HG_NODE: $HG_OLDPHASE -> $HG_PHASE" and - abort: pretxnclose-bookmark.force-forward hook exited with status 1 - [255] + abort: pretxnclose-bookmark.force-public hook exited with status 255 + [255] These failures originated in ee5f0d047b41::f6d17075608f.

test-log-linerange: open binary file in binary mode The '\n' characters were being translated on Windows, throwing off hg and git hashes, as well as the file content in the diff.

test-arbitraryfilectx: stabilize for Windows Previously, the second last test (context.arbitraryfilectx(..)) returned True on Windows. I changed the repo setup sequence to import a patch, so that way the repo would have a proper symlink. That made the last test fail, since it is comparing files in wdir(), one of which is not the expected symlink. Apparently the (feature !) line matching doesn't work well with (no-eol), so I had to conditionalize the test instead of the output.

dirstate: remove excess attribute lookups for dirstate.status (issue5714) A recent refactor added a layer of abstraction to the dirstate which makes doing things like 'foo in dirstate' now require some extra Python attribute lookups. This is causing a 100ms slow down in hg status for mozilla-central. The fix is to hoist the inner dict's functions onto the main class once the lazy loading it complete, as well as store the actual functions before doing the status loop (as is done for other such functions). In my testing, it seems to address the performance regression, but we'll need to see the perf run results to know for sure. Differential Revision: https://phab.mercurial-scm.org/D1257

dirstate: avoid reading the map when possible (issue5713) (issue5717) Before the recent refactor, we would not load the entire map until it was accessed. As part of the refactor, that got lost and even just trying to load the dirstate parents would load the whole map. This caused a perf regression (issue5713) and a regression with static http serving (issue5717). Making it lazy loaded again fixes both. Differential Revision: https://phab.mercurial-scm.org/D1253

dirstate: move clear onto dirstatemap class A future diff will move the lazy loading aspect of dirstate to the dirstatemap class. This means it requires a slightly different strategy of clearing than just reinstantiating the object (since just reinstantiating the object will lazily load the on disk data again later instead of remaining permanently empty). So let's give it it's own clear function. Differential Revision: https://phab.mercurial-scm.org/D1252

internal-doc: document the config register mechanism This explains the various usage and feature of the config register introduced in Mercurial 4.3 and 4.4.

help: clarify the pre-txnclose-phase documentation Gregory Szorc requested some clarification.

internal-doc: document the 'phases' parameters to 'getbundle' The getbundle wireprotocol method has some extended documentation. We update it with the next parameters introduced for binary phases.

status: update the help to indicate that clean files are not normally tersed The same applies to '?' if --quiet is used (or any of the other states if some of -marduic is specified), but I couldn't figure out how to express that clearly.

terse: split on repo separator instead of os.sep (issue5715) The paths being processed are from scmutil.status, and therefore normalized to '/' separators.

makefile: put format-c into .PHONY targets

statichttprepo: prevent loading dirstate over HTTP on node lookup (issue5717) This seems a bit hacky, but works well. There should be no reason that static-http repo had to load dirstate. Initially I tried to proxy os.stat() call through vfs so that statichttpvfs could hook it, but there wasn't a good error value which the statichttpvfs could return to get around the util.filestat issue.

test-static-http: show all files accessed over HTTP This provides some confidence on files that should be loadable over HTTP. Hopefully it will prevent future bugs.

server: drop executable bit from daemon log file The logfile option was unused, so it was okay until now.

setup: filter out devel-warn messages from system hg If we're going to use the user's installed and configured hg command (which we do since 8b20338b989e), we should prevent devel-warn messages from interfering with locating it.

setup: remove duplicate assignment of HGRCPATH HGRCPATH is already set in localhgenv().

packaging: update book URL in Mac and Windows READMEs

commands: be more uniform in an "hg update" error message "can only specify one of -C/--clean, -c/--check, or -m/merge" becomes "can only specify one of -C/--clean, -c/--check, or -m/--merge"

update: mention long options explicitly in description of merge.update() The short options "-c" and "-C" may be confusing for a novice reading the documentation. Let's try to be more explicit, also mentioning the equivalent long options ("--check" and "--clean") in the comments.

mq: copy pager attributes back to qrepo.ui If the legacy pager extension is enabled, a pager is started through repo.ui at dispatch._runcommand(). After that, mqcommand() creates a qrepo with a fresh repo.baseui, at which point pager information was lost and another pager would be spawned by the modern pager interface. This is a minimal workaround for the problem.

registrar: host "dynamicdefault" constant by configitem object This is the common pattern seen in the other registrar classes.

configitems: register 'ui.editor'

help: fix typo in hg merge documentation

Added signature for changeset 1e2454b60e59

Added tag 4.4-rc for changeset 1e2454b60e59

help: do not abort topicmatch() because of unimportable extensions This is alternative workaround to D1198, originally spotted by the earlier version of the releasenotes extension.

configitems: register 'email.to' and 'patchbomb.to'

exchange: propagate the subfunctions return The parts generator can return a callback to handle server reply. We should propagate the return for correctness.

merge default into stable for code freeze # no-check-commit because default contains new vendored code

log: add an assertion about fctx not being None in patch.diff() As noted in the comment, this should not happen as removed files (the cause of fctx2 being None) are caught earlier.

log: handle removed files with --line-range patterns I.e. abort when the pattern corresponds to a removed file, as done with bare file patterns.

log: disable bare file patterns with --line-range Currently, specifying both a line-range pattern and a bare file pattern results in an AND operation whereas we probably want an OR so that bare file patterns are like a line-range pattern with all lines specified. So, until this works as expected, we disable this.

log: switch to FROMLINE:TOLINE syntax for -L/--line-range This is more consistent with the followlines() revset.

log: handle binary files in --line-range patterns When a file is binary patch.trydiff() would yield None for 'hunkrange'. Handle this case in the hunksfilter() callback. Add tests with and without diff.git option as binary handling differs depending on this option's value.

build: build deb/rpm independently on config/extensions in the host system Reverts 5aac617a028d and replaces it with a more general solution. - works for both rpm and deb - sidesteps eventual problems with local extensions that have nothing to do with the build process (hg-git, for example, fails with version 4.4 because dedab036215d removed peerrepository, and hg-git still uses it as of 0.8.9)

config: also gather effect-flags on experimental.evolution Effect-flags config was in flight while the previous evolve config renaming was written. Now that both landed, gather effect-flags in experimental.evolution like the others evolve-related configurations. Differential Revision: https://phab.mercurial-scm.org/D1197

obsfate: rename obsfate into obsolete in default mapfile Like the previous patch, replace obsfate by obsolete in default mapfile. Differential Revision: https://phab.mercurial-scm.org/D1190

obsfate: rename obsfate into obsolete in changeset_printer Yuja's comment on the original obsfate about how we would translate obsfate and the recent discussions about exposing users to new concepts and names lead have led me to think that 'obsfate' should be treated as internal jargon. End- users should not be aware of obsfate, so we replace 'obsfate' by 'obsolete' in changeset_printer. It will be easier to understand for end-users, easier to translate and closer to the original Evolve obsfate output. I'm aware it's extremely late in the cycle but I think it's an UX improvement for the end-users. Differential Revision: https://phab.mercurial-scm.org/D1189

tests: add some (?) output lines to catch "helpful" output from Solaris diff Differential Revision: https://phab.mercurial-scm.org/D1196

sparse-read: ignore trailing empty revs in each read chunk An empty entry in the revlog may happen for two reasons: - when the file is empty, and the revlog stores a snapshot; - when there is a merge and both parents were identical. `hg debugindex -m | awk '$3=="0"{print}' | wc -l` gives 1917 of such entries in my clone of pypy, and 113 on my clone of mercurial. These empty revision may be located at the end of a sparse chain, and in some special cases may lead to read relatively large amounts of data for nothing.

fsmonitor: use nonnormalset from dirstatemap `dirstate._nonnormalset` has been moved to `dirstate._map.nonnormalset` by 60927b19ed65 (dirstate: move nonnormal and otherparent sets to dirstatemap) and is guaranteed to be existed. Let's update fsmonitor code to use the new `nonnormalset`. Thix fixed a perf regression that slows down `hg status` by 0.5 seconds in one of our production repos. Differential Revision: https://phab.mercurial-scm.org/D1184

registrar: don't i18n ProgrammingError message Differential Revision: https://phab.mercurial-scm.org/D1188

registrar: move "constant" possiblecmdtypes to class level While at it, switch to set literal syntax. Differential Revision: https://phab.mercurial-scm.org/D1187

tests: de-flake test-run-tests.t's "--jobs=2 --first" test Once in a while, test-nothing.t finished before test-failure.t or test-failure-copy.t (I haven't checked which one actually gets run first). Since there already are two tests that will fail in the same way, just run those two instead so the diff will be reproducible and not timing-dependent. The test case was added in 9a20f53e436f (run-tests: handle --jobs and --first gracefully, 2014-10-09), and I have checked that backing that out results in two failures being printed. Note that the summary may still include multiple tests even if --first is given, it's just that the diff is only printed for the first failure. Differential Revision: https://phab.mercurial-scm.org/D1186

cmdutil: fix status tersing on Python 3 Differential Revision: https://phab.mercurial-scm.org/D1183

hghave: fix clang-format check to use bytes regex Fixes hghave on Python 3. Differential Revision: https://phab.mercurial-scm.org/D1182

configitems: make all regular expressions bytes and not native str Fixes many tests on Python 3. Differential Revision: https://phab.mercurial-scm.org/D1181

convert: register missed subversion config items Should fix the build failures on the Solaris builders. Differential Revision: https://phab.mercurial-scm.org/D1180

amend: error out if the note is greater than 255bytes In obsmarker we can't store keys and values of more than 255 bytes in metadata. If we try to do so, ProgrammingError is raised. The note flag to amend stores the note in obsmetadata. If a user will try to store a larger note, he will encounter ProgrammingError which is wrong. We must error out early. Thanks to Yuya for warning about this. Differential Revision: https://phab.mercurial-scm.org/D1179

show: move configitems to core chgserver.py is also checking the config and will get: devel-warn: accessing unregistered config item: 'commands.show.aliasprefix' at: mercurial/chgserver.py:109 if the config is not registered. Differential Revision: https://phab.mercurial-scm.org/D1178

chgserver: do not treat HG as sensitive environ when CHGHG is set When `$CHGHG` is set, `$HG` is ignored by the chg client. Removing it from chg's sensitive environment list would avoid starting up servers unnecessarily when `$CHGHG` is the same while `$HG` is different. Differential Revision: https://phab.mercurial-scm.org/D1177

fsmonitor: declare missing config options These were added in 4aa57627692a. Attempting to run the test harness with fsmonitor enabled spews a whole bunch of devel warnings due to these options not be declared. Differential Revision: https://phab.mercurial-scm.org/D1176

fsmonitor: warn when fsmonitor could be used fsmonitor can significantly speed up operations on large working directories. But fsmonitor isn't enabled by default, so naive users may not realize there is a potential to make Mercurial faster. This commit introduces a warning to working directory updates when fsmonitor could be used. The following conditions must be met: * Working directory is previously empty * New working directory adds >= N files (currently 50,000) * Running on Linux or MacOS * fsmonitor not enabled * Warning not disabled via config override Because of the empty working directory restriction, most users will only see this warning during `hg clone` (assuming very few users actually do an `hg up null`). The addition of a warning may be considered a BC change. However, clone has printed warnings before. Until recently, Mercurial printed a warning with the server's certificate fingerprint when it wasn't explicitly trusted for example. The warning goes to stderr. So it shouldn't interfere with scripts parsing meaningful output. The OS restriction was on the advice of Facebook engineers, who only feel confident with watchman's stability on the supported platforms. .. feature:: Print warning when fsmonitor isn't being used on a large repository Differential Revision: https://phab.mercurial-scm.org/D894